The Type 1 Communications Adapter that is built into my IBM 3174 controller can operate in a number of ways, using one of three different serial communications links - EIA (RS 232), X.21 and V.35 at the physical level and BSC or SDLC at the link protocol level.
The administrator of the 3174 configures the controller to specify what version of the physical and link layers are to be used. The microcode will check to be sure that the connected hardware is consistent with the configuration, using a few of the signal lines on the DB25 connector to interrogate the attached cable type.
The cables all have a switch, Oper/Test, that can wrap signals back to allow the microcode in the 3174 to validate the correct function of the Type 1 Communications Adapter and the rest of the controller. However, there are also Wrap Test plugs that make these same test connections along with minor differences that allow the microcode to recognize them as test plugs.
IBM has defined three wrap test plugs, one each for EIA, V.35 and X.21 physical layers. The wiring diagrams specify most of the wiring needed in these plugs but a few points are ambiguous. In one place it states that two wires are connected to each other, but not whether they are also tied to the signal ground line. Thus, I might have to make a few variants of the wrap test plugs until they are recognized as valid and work properly.
CHOSE A BREAKOUT PLUG TO ALLOW QUICK PROTOTYPING CHANGES
I could have used female DB-25 connectors and wired pins together to match the wiring diagrams. It might have required some rework if the ambiguities caused me to choose wrongly. To build a full set, at least three connectors have to be wired even if the uncertainties don't cause problems.
Instead, I chose a breakout board for the female DB-25 connector. This has all 25 pins routed out to terminal blocks, where I can insert and change wires rapidly to change the connections.
I will use this to build each of the three wrap-test plug types and experiment until it works properly.
Sorry, but I never really got a good sense of the point of this project. It occurs to me that if what you're trying for is a working 3270 terminal, that you might be better served by getting a 3276. A 3276 is basically a 3278 terminal with a built-in 3274 control unit. From the outside it looks almost identical to a 3278, but there's a panel on the backside that has 7 or 8 coax connectors for hooking up additional 3278 terminals.
ReplyDeleteBut maybe you already knew this, and you have something else in mind as your end goal. In any event, you probably don't want to give up after so much research and work thus far. Thanks for sharing!
Hi Gene.
ReplyDeleteThe point I am at currently is where I would be if I did find and restore a 3276. It provides a controller inside with a communications adapter and would need a far end connected to it before the terminal would work.
The other factor is availability. There are quite a few of the 317x products which replaced the 327x equivalents but are a decade younger. 3178 and 3179 terminals are relatively affordable and easily found, while 3278 and 3276 are quite a bit harder to find and commensurately more expensive.
there is one 3276 on Ebay right now, but from the listing there is a box of the loose cards from the controller portion and no indication that the CRT even works. At the right price it might be an interesting restoration target but right now it is pricey and high risk due to lack of evidence that it can work again.
I also have Alexander Kay's 3174 replacement working - shields that sit on an Arduino, connect by coax to the terminals, and talk with Python code on a PC to communicate to Hercules or other mainframe instances.
These projects are my retirement hobby interests, so they don't need much of a point except that they capture my interest and keep me occupied.